Parvoviruses are small, single-stranded, non-enveloped DNA viruses between twenty to thirty nanometers in diameter. The genomes of parvoviruses are approximately 5000 nucleotides long, containing two open reading frames. The left-hand open reading frame codes for the proteins responsible for replication (Rep), while the right-hand open reading frame encodes the structural proteins of the capsid (Cap). All parvoviruses have virions with icosahedral symmetry composed of a major Cap protein, usually the smallest of the Cap proteins, and one or two minor Cap proteins. The Cap proteins are generated from a single gene that initiates translation from
Most parvoviruses have narrow host ranges; the tropism of B19 is for human erythroid cells (Munshi et al., (1993) J. Virology 67:562), while canine parvovirus has a tropism for lymphocytes in adult dogs (Parrish et al., (1988) Virology 166:293; Chang et al., (1992) J. Virology 66:6858). Adeno-associated virus, on the other hand, can replicate well in canine, mouse, chicken, bovine, monkey, as well as numerous human lines, when the appropriate helper virus is present. In the absence of helper virus, AAV will infect and establish latency in all of these cell types, suggesting that the AAV receptor is common and conserved among species. Several serotypes of AAV have-been identified, including serotypes 1, 2, 3, 4, 5 and 6.
Adeno-associated virus (AAV) is a dependent parvovirus twenty nanometers in size which requires co-infection with another virus (either adenovirus or certain members of the herpes virus group) to undergo a productive infection in cultured cells. In the absence of co-infection with helper virus, the AAV virion binds to a cellular receptor and enters the cell, migrating to the nucleus, and delivers a single-stranded DNA genome that can establish latency by integration into the host chromosome. The interest in AAV as a vector has centered around the biology of this virus. In addition to its unique life-cycle, AAV has a broad host range for infectivity (human, mouse, monkey, dog, etc.), is ubiquitous in humans, and is completely nonpathogenic. The finite packaging capacity of this virus (4.5 kb) has restricted the use of this vector in the past to small genes or cDNAs. To advance the prospects of AAV gene delivery, vectors sufficient to carry larger genes must be developed. In addition, virions that specifically and efficiently target defined cell types without transducing others will be required for clinical application.
The capsid proteins of AAV2 are Vp1, 2, and 3 with molecular weights of 87, 73, and 62 kDa, respectively. Vp3 represents nearly 80% of the total protein in intact virions, while Vp1 and Vp2 represent 10% each (Muzyczka, (1992) Curr. Topics Microbiol. Immunol. 158:97; Rolling et al., (1995) Molec. Biotech. 3:9; Wistuba et al. (1997) J. Virology 71:1341). Early studies of AAV2 support that all three capsid subunits are required to extract single stranded genomes from the pool of replicating double stranded DNA. These genomes are then sequestered into preformed immature particles that maturate to infectious particles. These particles have a density between 1.32 and 1.41 g/mL in cesium chloride and sediment between 60S and 125S in sucrose (Myers et al., (1981) J. Biological Chem. 256:567; Myers et al., (1980) J. Virology 35:65).
Previous mutagenesis studies of AAV2 capsids have shown that insertions and deletions in the Vp3 domain completely inhibit the accumulation of single stranded virions and production of infectious particles (Hermonat et al., (1984) J. Virology 51:329; Ruffing et al., (1992) J. Virology 66:6922). Yang et al., (1998) Human Gene Therapy 9:1929, have reported the insertion of a sequence encoding the variable region of a single chain antibody against human CD34 at the 5′ end of the AAV2 Vp1, Vp2 or Vp3 coding regions. These investigators observed extremely low transduction of CD34 expressing KG-1 cells by AAV virions containing the Vp2 fusion protein (1.9 transducing units/ml or less, sentence spanning pages 1934–35). KG-1 cells are reportedly not permissive to infection by a wild-type rAAV vector. These results with the Vp2 fusion AAV are suspect as transduction of KG-1 cells by this virus was essentially insensitive to an anti-AAV capsid antibody (430 vs. 310 transducing units/ml; Table 2), whereas transduction of HeLa cells was markedly reduced by this antibody (63,2000 vs. <200 transducing units/ml; Table 2). No characterization of the putative fusion virions was undertaken to confirm that the particles contained the Vp2 fusion protein, the antibody was expressed on the capsid surface, or that the particles bound CD34 proteins. In addition, rAAV particles could only be produced if all three wild-type capsid subunits were provided, in addition to the chimeric subunit (Page 1934, Col. 2, lines 5–12). Collectively, these results suggest the chimeric subunits were not incorporated into viable AAV particles, and the low level of chimeric protein observed in target cells was, in fact, due to cellular uptake of chimeric capsid protein or protein aggregates by other mechanisms.
Several studies have demonstrated that parvovirus capsid proteins can be mutated and virion assembly studied. In one study, the coding region for 147 amino acids of the hen egg white lysozyme was substituted for B19 Vp1 unique coding sequence. This modification resulted in purified empty particles that retained lysozyme enzymatic activity (Miyamura et al., (1994) Proc. Nat. Acad. Sci. USA 91:8507). In addition, expression of peptides (9 and 13 residues) in B19 Vp2 resulted in empty particles that were immunogenic in mice supporting surface presentation of the insertions (Brown et al., (1994) Virology 198:477). In a more recent study, the CD8+CTL epitope (residues 118–132) against lymphocytic choriomeningitis virus (LCAAV) nucleoprotein was inserted into the Vp2 capsid protein of porcine parvovirus (ppv) (Sedlik et al., (1997) Proc. Nat. Acad. Sci. USA 94:7503). This capsid protein, with the epitope cloned at the N-terminus, self-assembled when expressed in a baculovirus system. This chimeric virus-like particle was then used to immunize mice against a lethal challenge from LCAAV. While these studies evaluated capsid structure and assembly, they did not address the issue of packaging B19 genomes into the altered capsids.
Recombinant (r)AAV vectors require only the inverted terminal repeat sequences in cis of the 4679 bases to generate virus. All other viral sequences are dispensable and may be supplied in trans (Muzyczka, (1992) Curr. Topics Microbiol. Immunol. 158:97). Attractive characteristics of AAV vectors for gene therapy are the stability, genetic simplicity, and ease of genetic manipulation of this virus. While each of these factors remains valid, some obstacles to the application of rAAV vectors have recently come to light. These include inefficiency of vector transduction and packaging constraints. It is not surprising, given the cryptic nature of this virus, that new insights into its biology have surfaced only after extensive research with rAAV vectors, which are more easily assayed compared with wild-type AAV.
With respect to the efficiency of vector transduction, several recent studies have shown great promise in terms of duration of transgene expression in vivo; however, there has been a shortfall in the efficiency of transduction, which was unexpected based on previous results in vitro (Flotte et al., (1993) Proc. Nat. Acad. Sci. USA 90:10613). One of the first experiments in rodents to demonstrate the utility of rAAV vectors in vivo was aimed at transduction of brain tissue in rat (Kaplitt et al., (1994) Nature Genet. 7:148). In addition to brain, muscle has been found to be efficiently transduced in vivo by AAV vectors, demonstrating long term gene expression (at least 1.5 years), lack of immune response, and no vector toxicity (Xiao et al., (1996) J. Virol. 70:8098; Clark et al., (1996) Hum. Gene Ther. 8:659; Fisher et at, (1997) Nat Med. 3:306; Monahan et al., (1998) Gene Ther. 5:40). The primary steps that influence efficient vector delivery are virus entry and conversion of second strand synthesis (see Ferrari et al., (1996) J. Virology 70:3227–34).
The overall success of AAV as a general-purpose viral vector depends on the ability to package larger than full-length AAV genomes (5 kb) into rAAV vectors. Studies by Dong et al., (1996) Hum. Gene Ther. 7:2101, have determined the packaging limitations using rAAV vectors as between 104% and 108%. This packaging restriction precludes the use of a number of important genes currently being tested for human gene therapy (e.g., the dystrophin gene or current mini-dystrophin constructs).
Accordingly, there remains a need in the art for improved virus vectors with greater packaging limits and transduction efficiency than AAV vectors. In addition, there remains a need for virus vectors with altered tropisms as compared with AAV vectors.